private void processCoreferencedEntities(Integer docKey, List<Entity> entities)
throws SQLException {
Set<String> values = new HashSet<>();
Set<String> externalIds = new HashSet<>();
Set<String> geoJsons = new HashSet<>();
Class<? extends Entity> type = null;
for (Entity e : entities) {
values.add(e.getValue());
externalIds.add(e.getExternalId());
if (e instanceof Location) {
Location l = (Location) e;
try {
geoJsons.add(addCrsToGeoJSON(l.getGeoJson()));
} catch (BaleenException ex) {
getMonitor().warn("Unable to add CRS to GeoJSON", ex);
}
}
type = getSuperclass(type, e.getClass());
}
Integer entityKey = executeEntityInsert(docKey, values, externalIds, type.getName());
if (entityKey != null) {
for (String geoJson : geoJsons) {
executeEntityGeoInsert(entityKey, geoJson);
}
}
}
private List<String> getEntityIds(FSArray entityArray) {
List<String> entities = new ArrayList<>();
for (int x = 0; x < entityArray.size(); x++) {
Entity ent = (Entity) entityArray.get(x);
entities.add(ent.getExternalId());
}
return entities;
}
/**
* Convert from an entity to a map.
*
* @param entity the entity to convert
* @return a map containing the entity's fields (and history is required)
*/
public Map<String, Object> convertEntity(Entity entity) {
Map<String, Object> map = Maps.newHashMap();
convertFeatures(map, entity);
if (outputHistory && documentHistory != null) {
Collection<HistoryEvent> events = documentHistory.getHistory(entity.getInternalId());
convertHistory(map, events, entity.getInternalId());
}
putIfExists(map, fields.getExternalId(), entity.getExternalId());
return map;
}
private void processEntities(JCas jCas, Integer docKey) throws SQLException {
// Insert entities
Map<ReferenceTarget, List<Entity>> coreferenceEntities = new HashMap<>();
for (Entity ent : JCasUtil.select(jCas, Entity.class)) {
ReferenceTarget rt = ent.getReferent();
if (rt == null) {
rt = new ReferenceTarget(jCas);
}
List<Entity> entities = coreferenceEntities.getOrDefault(rt, new ArrayList<>());
entities.add(ent);
coreferenceEntities.put(rt, entities);
}
for (List<Entity> entities : coreferenceEntities.values()) {
processCoreferencedEntities(docKey, entities);
}
}
@Override
public void doProcess(JCas aJCas) throws AnalysisEngineProcessException {
List<Entity> toRemove = new ArrayList<Entity>();
FSIterator<Annotation> iter = aJCas.getAnnotationIndex(Entity.type).iterator();
while (iter.hasNext()) {
Entity e = (Entity) iter.next();
if (e.getConfidence() < confidenceThreshold
&& (!ignoreZeroConfidence || e.getConfidence() > 0.0)) {
toRemove.add(e);
getMonitor()
.debug(
"Low confidence entity found (ID: {}) - this entity will be removed",
e.getInternalId());
}
}
removeFromJCasIndex(toRemove);
}